Synopsis
Calculate and interpret common measures of disease frequency, association, diagnostic accuracy and intervention effect while recognising bias, confounding, uncertainty and population context.
- Incidence measures new events among people at risk over time; cumulative incidence uses a starting population at risk, whereas incidence rate uses person-time.
- Prevalence is the proportion with a condition at a point or during a period and depends on both incidence and duration; it is useful for service need but not usually causal sequence.
- Risk ratio compares risks, rate ratio compares incidence rates, and odds ratio compares odds; the odds ratio approximates the risk ratio only when the outcome is uncommon.
Reasoning priorities
Confirm the measure’s population, event and time basis.
State who can enter each count, who is at risk, start and stop times and whether repeat events are permitted.
Worked reasoning
Among 1,000 people, 100 truly have disease. A test is positive in 90 with disease and 90 without disease.
- Construct the table: true positive 90, false negative 10, false positive 90 and true negative 810; verify all four cells sum to 1,000.
- Calculate sensitivity 90/(90+10)=90% and specificity 810/(810+90)=90%; these condition on disease status.
- Calculate the final predictive answers: PPV 90/(90+90)=50% and NPV 810/(810+10)=98.8%, rounded to one decimal place.
- Verify interpretation: half of positive results are false positives in this 10%-prevalence population; recalculate rather than carrying PPV unchanged into a lower-prevalence setting.